An ontology-based approach for preprocessing in machine learning

Increasing pressure on internationally operating companies leads to the application of novel technologies, e.g., Machine Learning models. However, Machine Learning algorithms require preprocessing, i.e., data cleaning, which is time consuming and requires domain-specific knowledge. Formalized knowledge bases capture such domain-specific knowledge in a computer-interpretable way and have the potential to reduce manual efforts for this process. This paper presents a framework for semantic preprocessing, which is evaluated at the example of an industrial use case from the semiconductor industry.